Aerosol Science and Technology 34: 373 380 (2001)

c 2001 American Association for Aerosol Research

Published by Taylor and Francis

0278-6826=01=$12.00 C .00

Design and Performance Evaluation of a Novel

Double Cyclone

Y. Zhu,1 M. C. Kim,1 K. W. Lee,1 Y. O. Park,2 and M. R. Kuhlman3

1
Department of Environmental Science and Engineering, Kwangju Institute of Science and Technology,
Kwangju 500-712, Korea
2
Energy and Environmental Research Department, Korea Institute of Energy Research,
Taejon 305-343, Korea
3
Aerosol Science and Technology Assessment, Columbus, Ohio

the particles entrained in the gas, throwing them to the cyclone

A novel cyclone design, named double cyclone, adds an extra
cylinder wall into a conventional reversed ow cyclone. The phys-
ical performance of the double cyclone has been experimentally
compared to a widely used conventional cyclone by measuring the
particle concentrations both upstream and downstream of each cy-
clone with an Aerosizer. Tests with monodisperse polystyrene latex
(PSL) particles with diameters ranging from 0.60 m to 8.8 m
have shown that the double cyclone provides higher collection ef-
ciency than the conventional cyclone for a tested ow rate range
of 10 40 L/min. In addition, the collection ef ciency difference be-
tween the double cyclone and the conventional cyclone becomes
even larger as the air ow rate increases. Sharper particle-size sep-
aration characteristics were also observed for the double cyclone.
Furthermore, it was found that the double cyclone operates at a
lower pressure drop than the conventional cyclone at the same
normalized cut size.
INTRODUCTION
Cyclones, one of the most widely used industrial dust col-
lectors, are very rugged in design, reliable in performance, and
easy to maintain. Over the years, many different types of cy-
clones have been built (Figure 1). In a reversed ow cyclone,
or cylinder and cone design with a tangential inlet as shown in
Figure 1a, aerosol enters the cyclone at the cylinder top, where
the con guration of the entry causes the gas to spin forming a
vortex. The high tangential velocity exerts a centrifugal force on
Received 1 June 1999; accepted 10 November 1999.
This work was performed as part of the research project
titled Development of High Ef ciency Dust Precharger and Mem-
brane (Project No. 1997C-CC020-P-02) sponsored by Korea Research
and Development Management Center for Energy and Resources.
Address correspondence to Ken W. Lee, Department of Envi-
ronmental Science and Engineering, Kwangju Institute of Science
and Technology, 1 Oryong-doug , Puk-gu, Kwangju 500-712, Korea.
E-mail: lee@env.kjist.ac.kr
wall for collection. Below the bottom of the gas exit tube, the
spinning gas gradually migrates inward to a central core along
the cyclone axis and, from there up, nally out through the gas
exit. Dust collected at the wall descends to the collection trap at
the bottom of the cone, primarily due to the downward compo-
nent of the gas velocity at the cyclone wall rather than to gravity.
Although the reversed ow cyclone is the most popular cy-
clone design, several limitations are inherent in its design. First,
the collection ef ciency of ne particles is not high enough
when a reversed- ow cyclone is operated at ordinary ow rates.
Generally, the cut size of conventional cyclones is about 5 m
(Dirgo and Leith 1985; Iozia and Leith 1989). Second, cyclones
generally do not provide sharp particle-size separation between
the collected and uncollected particles (Willeke et al. 1998). In
addition, when the turbulent rotational ow migrates up in the
cone, the separated dust has the possibility of escaping from the
exit tube with the upward rotational gas ow (Ogawa 1984).
The axial ow cyclone is another frequently used cyclone
design. Figure 1b shows a typical once through axial ow cy-
clone. The rotational ow is created by the guide vanes (Daniels
1957). Centrifugal force acts as the particle removal power and
clean gas ows out through the annular space. Since there is no
reversed ow in this design, the possibility of particle reentry
due to the upward vortex is diminished. However, the collection
ef ciency is generally lower than the tangential inlet cyclones
because of the relatively weak rotational gas ow achieved by
means of the guide vanes.
Figure 1c shows a reversed axial ow cyclone in which the
dust laden air ows axially into the cyclone. After it passes
through the guide vane, it rotates to form a vortex. Again, the
particles are thrown to the wall by centrifugal force. The clean
air escapes from the exit tube. Both types of axial ow cyclones
are used in multicyclone banks for gas cleaning.
Since cyclones have been used extensively in industry, both
theoretical and experimental studies on cyclones were carried

373
374 Y. ZHU ET AL.
Figure 1. Schematics of different cyclone designs: (a) reversed ow cyclone; (b) once through axial ow cyclon; (c) reversed
axial ow cyclone; gas trajectory; - - - - - particle trajectory.
out from the early 1940s to the present (Stairmand 1951;
Dietz 1981; Boysan et al. 1982; Dirgo and Leith 1985; Lee et al.
1985; Iozia and Leith 1989; Li and Wang 1989; Iozia and Leith
1990; Kim and Lee 1990; Coker 1993; Liden and Gudmundsson
1997; Zhu and Lee 1999). Although our knowledge of ow and
particle capture mechanisms operating inside a cyclone has in-
creased over the years, the exact mechanisms of removing parti-
cles are still not fully understood. Thus designs of cyclones still
rely largely upon experience, trial, and design guides.
In this study, a new cyclone design is introduced in an effort to
increase the particle collection ef ciency and to overcome some
of the limitations of conventional cyclones. In the new design,
an additional cylinder wall was introduced into the cyclone body
which separates the total inner space into two annular sections.
One is between the cyclone wall and the cylinder wall and the
other is between the cylinder wall and a removable exit tube,
which exits downward rather than upwards, as in a regular cy-
clone design. Due to this design feature, entrained air is forced to
ow one more vortex inside the cyclone body and particles may
be collected both on the cyclone wall and on the cylinder wall.
Deposited particles descend to the lower part of the device and
may be collected by removing the exit tube. This device has the
appearance of two cyclones combined coaxially together and is
called the double cyclone.
The objective of the present study is to compare the perfor-
mance of the double cyclone with that of a conventional cyclone.
Since the centrifugal force plays a very important role in cyclone
performance and since one more vortex is introduced in the dou-
ble cyclone, the double cyclone is expected to provide a higher
collection ef ciency than that of the conventional cyclone and
thus work as a better aerosol sampling device.
EXPERIMENTAL
Double Cyclone
A prototype of the double cyclone was constructed out of
glass. Since there is no theory that is suf ciently accurate to serve
as a basis for small cyclone design, the dimensions of the double
cyclone were selected to be comparable to those of conventional
cyclones that had been previously evaluated. Figure 2 shows
the detailed design dimensions (mm) of the double cyclone.
The particle laden air enters the double cyclone tangentially
on the top part of the cylinder and forms a downward vortex
along the cyclone wall surface. When the vortex reaches its end,
air then ows upward in the annular space between the exit
tube, which faces down, and the cylinder wall to the top of the
cyclone. Since there is no way for air to escape at the apex
of the cone, it nally ows down again and escapes through
the exit tube. Thus instead of the traditional two vortexes, the
new design incorporates three vortexes in one cyclone design.
Particles may be collected on the surface of both cyclone wall
and cylinder wall due to the strong centrifugal force generated
by the vortexes. Since the double cyclone is intended to be used
as a small sampling device, not too much aerosol mass load
is expected and thus no grit pot is introduced in the design.
This design has a signi cant advantage. Due to the additional
vortex, centrifugal force is strengthened and the residence time
inside the cyclone body is lengthened. Both contribute to a higher
collection ef ciency.
The evaluation of most air sampling devices is not straight-
forward, and therefore the utility of new instruments is fre-
quently assessed by comparative tests with widely used and well-
characterized standard samplers. In this work, the performance
 A NOVEL DOUBLE CYCLONE
Figure 2. Schematic of the double cyclone (above dimensions
are all in mm).
of the double cyclone is evaluated by comparing the particle col-
lection ef ciency of double cyclone with that of a conventional
cyclone, namely, a high ef ciency Stairmand cyclone.
High Efciency Stairmand Cyclone
The performance of the double cyclone has been compared
with that of the widely used glass high ef ciency Stairmand cy-
clone, which has been used as a reference sampler for collecting
airborne particles since the 1950s (Stairmand 1951). This kind
of cyclone (shown in Figure 1a) is a typical reversed ow or
cone-under-cylinde r design. The performance of the high ef -
ciency Stairmand cyclone has been studied extensively through
both laboratory and eld works (Dietz 1981; Boysan et al. 1982;
Dirgo and Leith 1985; Lee et al. 1985; Iozia and Leith 1989; Li
and Wang 1989; Iozia and Leith 1990; Kim and Lee 1990; Coker
1993; Liden and Gudmundsson 1997; Zhu and Lee 1999). Sev-
eral other cyclone designs have been evaluated by comparing
their performance with that of the high ef ciency Stairmand
cyclone (Razgaitis and Gusenther 1981; Jaroszczyk and Ptak
1985; Moore and McFarland 1990). The dimensions of the high
ef ciency Stairmand cyclone used in this study are shown in
Figure 3.
375
Figure 3. Schematic of the test high ef ciency Stairmand cy-
clone (above dimensions are in mm).
Experimental Setup
A diagram of the experimental system is shown in Figure 4.
The main system components consist of an aerosol generator, a
test cyclone, and an aerosol detector. Monodisperse polystyrene
latex (PSL) particles (Bangs Laboratories, Carmel, IN) were
employed for all experiments. Aerosols containing PSL particles
ranging from 0.60 m to 8.8 m in diameter were produced by
an atomizer (TSI, Model 9302). The material density of PSL
is 1.05 g/cm3 . A drying chamber was placed in-line after the
aerosol generator so that the liquid in the droplets containing
a PSL particle was evaporated completely before the particle
entered the particle detector. By doing so, the PSL particles were
clearly distinguished from the residue particles of the droplets
that did not contain any PSL particle (Grinshpun et al. 1997).
All the experiments were carried out at room temperature and
atmospheric pressure.
The experimental procedures for evaluating the particle col-
lection ef ciencies of these cyclones follow. After particles of
a desired size were generated, the particle-laden air was drawn
376 Y. ZHU ET AL.

Figure 4. Schematic diagram of experimental system setup.

through a test cyclone at a desired ow rate using a vacuum
pump. The ow rates employed in this study ranged from
10 L/min to 40 L/min in 10 L/min increments. Flow through the
cyclone was measured by a rotameter that was calibrated against
a wet test meter. Pressure drop across the cyclone was measured
by a Megnehelic gauge (Dwyer Instruments Inc., Michigan City,
IN). The tested particles were sampled alternately through one
of two sampling lines: one upstream and the other downstream
of the test cyclone. Since both sampling lines were identical,
any losses that may have occurred in the sampling lines were
assumed to be the same. In this study, an Aerosizer/Diluter
(Amherst Process Instruments, Inc., Amherst, MA) combination
was used to measure the aerosol concentrations both upstream,
C u , and downstream, C d , of the test cyclone. The collection ef -
ciency, E c , of the test cyclone was determined through measure-
ments of the upstream and downstream aerosol concentrations:
Cu Cd Cd
Ec D D1 : [1]
Cu Cu
diameter downstream of the cyclone. Therefore the effect of
swirling ow on particle concentration measurements was not
expected to be signi cant. To obtain statistically signi cant re-
sults, for each case (cyclone dimension, ow rate, and particle
diameter), ve replicate experiments were conducted. After the
highest and the lowest measurement ef ciency data were dis-
carded, the three remaining data sets were averaged to obtain
the particle collection ef ciency. Throughout the experiments,
it was found that the reproducibility of the data was with ap-
proximately 5% agreement.
RESULTS AND DISCUSSION
To compare the performance of the double cyclone and the
conventional cyclone, Figure 5 shows the particle collection ef-
ciency versus particle diameter for both cyclones at 10 L/min
and 40 L/min. As indicated by many researchers (Dietz 1981;
Boysan et al. 1982; Dirgo and Leith 1985; Iozia and Leith 1989,

The Aerosizer was interfaced with a personal computer ca-

pable of measuring not only the total concentration, but also the
particle size distribution over the desired size ranges. The in-
strument did not have any coincidence losses over the measured
aerosol concentration range of 0.1 300 particles/cm3 (Ulevicius
et al. 1997). It was also noted that the PSL particle size in-
dicated by the Aerosizer was in excellent agreement with the
nominal PSL particle size value. The sampling ow rate of
1.0 L/min drawn by Aerosizer was taken into account when
adjusting the ow rate through the cyclone. For large cyclones,
a simple sampling probe may not measure the particle concen-
tration accurately in the cyclone exit tube because the swirling
motion of the air tends to decrease the uniformity of particle
concentration distribution. However, in the present experiment
the cyclone and the exit tube were relatively small. In addi- Figure 5. Particle collection ef ciencies of the double cyclone
tion, the sampling point was located about eight times the tube and the Stairmand cyclone tested at 10 L/min and 40 L/min.
 A NOVEL DOUBLE CYCLONE
1990; Kim and Lee 1990; Coker 1993; Liden and Gudmundsson
1997; Zhu and Lee 1999), the collection ef ciencies, shown in
Figure 5, increased signi cantly with increasing particle size or
gas ow rate. All the measurement data of 10 L/min also are well
described by the so-called S shape curve. Figure 5 indicates
that at a xed ow rate of 10 L/min, the ef ciency curve of the
double cyclone is seen to be to the left of that of the conventional
cyclone, indicating a higher collection ef ciency or a lower 50%
cut size (d50 ). It is also noted that, as the ow rate increases
(40 L/min), the S shape curve of the double cyclone moves
further left of the conventional cyclone, which implies an even
higher collection ef ciency.
In order to systematically compare the performance of the
double cyclone and the conventional cyclone, Figure 6 was gen-
erated to depict cut sizes versus all tested ow rates for both the
double cyclone and the conventional cyclone. For all four ow
rates, the cut sizes for the double cyclone were consistently much
smaller than those of the conventional design. In addition, the
difference became more and more signi cant as the ow rate
increased, indicating that the double cyclone provided higher
collection ef ciency than the conventional cyclone, especially
at relatively high ow rates.
Most recently, researchers found that d50 was not the best
indicator for evaluating cyclone performance. They found that
a smaller cyclone tended to have a higher collection ef ciency
than a larger cyclone with the same dimensional design ratios.
To compare cyclones with different dimensions, d50 =Dc (where
Dc is cyclone body diameter) may be a better parameter to use
(Moore and McFarland 1993, 1995; Kenny and Gussman 1997;
Liden and Gudmundsson 1997). Since the body diameters for
the double cyclone and the test Stairmand cyclone were not iden-
tical in this study, Figure 7 was generated to plot normalized cut
sizes, d50 =Dc, versus ow rates. To evaluate the performance of
the double cyclone in a comprehensive way, data from several
previously published experimental studies were included. Ex-
cept for the lowest ow rate region (Kenny and Gussman 1996)
Figure 6. Comparison of cut sizes for the double cyclone and
the conventional cyclone at different ow rates.
377
and the highest ow rate region (Zhu and Lee 1999), the double
cyclone provided smaller normalized cut sizes than other con-
ventional cyclones in the tested ow rate region. As mentioned
before, cyclones generally have dif culty in providing a sharp
particle-size separation. In fact, the sampling effectiveness of a
cyclone is characterized not only by the normalized cut size, but
also by a slope which is the square root of the ratio of the diam-
eter of particles removed by the cyclone with an 84% ef ciency
(d84 ) to the diameter removed with a 16% ef ciency (d16 ) as
shown in Equation (2).
s
d84
Sc D [2]
d16
A slope of 1 indicates a step function, while a slope that ex-
ceeds 2 does not provide a de nitive size cut slope. Slopes of
1.3 1.5 are considered to provide a well-de ned particle size
separation (Watson and Chow 1993). The slopes of both the
double cyclone and the test conventional cyclone at different
ow rates were calculated using Equation (2) and are listed in
Table 1. According to Table 1 the slopes of the double cyclone
were always smaller than those of the conventional test cyclone,
which indicates a sharper particle-size separation is achieved.
Thus at a xed ow rate, the double cyclone provided a smaller
normalized cut size and a better designed particle-size separa-
tion curve. Both of these features contribute to a more ef cient
cyclone design.
Another important issue in cyclone design is energy con-
sumption, which may be indirectly evaluated by examining the
pressure drop across the cyclone (Jaroszczyk and Prak 1985;
Kim and Lee 1990; Grif ths and Boysan 1996; Zhu and Lee
1999). Generally speaking, the pressure drop, 1Pc , in a cyclone
can be de ned as the pressure difference between the mean total
pressure (static pressure plus dynamic pressure) in the inlet, Pi T ,
and the mean total pressure in the exit tube, PeT , as follows:
1Pc D Pi T PeT
V2 V 2 C Ver2
D Pis C i Pes C e ; [3]
2 2
where Pis is the static pressure in the inlet tube, Pes is the static
pressure in the exit tube, is the gas density, Vi is the inlet
velocity, (Vi2 =2) is the dynamic pressure in the inlet tube,
Ve is the tangential velocity of gas in the exit tube, Ver is the
axial velocity of gas in the exit tube, Ve2 C Ver2 is the square of
the mean velocity in the exit tube, and ((Ve2 C Ver2 )=2) is the
dynamic pressure in the exit tube.
It is almost impossible to exactly measure Vi ; Ve , and Ver
in such a small cyclone design. Thus the mean velocity in the
inlet tube, Vi , and in the exit tube, Ve , is assumed to be the
ratio of the ow rate passing through the cyclone to the cross-
section area of the inlet tube and the exit tube, respectively.
378 Y. ZHU ET AL.

Figure 7. Flow rates versus normalized cut sizes for different cyclones.

Equation (4) follows: observed by prior researchers, the pressure drop became larger
as the ow rate is increased (Jaroszczyk and Ptak 1985; Kim
Q Q
Vi D and Ve D ; [4] and Lee 1990: Grif ths and Boysan 1996; Zhu and Lee 1999).
Ai Ae As expected, at a xed ow rate, the pressure drop across the
where Q is the ow rate passing through the cyclone, Ai is double cyclone was greater than that across the conventional
the cross-sectional area of the inlet tube, and A e is the cross- cyclone. This occurs because the pressure drop is mainly due to
sectional area of the exit tube. Therefore Equation (3) is reduced the formation of the vortex, thus an enhanced rotational ow or
to Equation (5) as follows:

1Pc D Pi T PeT Table 1

Slopes of particle-size separation characteristics for the
(Q=Ai )2 (Q= Ae )2
D Pis C Pes double cyclone and the conventional test cyclone at
2 2 different ow rates

Q2 1 1
D ( Pis Pes ) C 2 : [5] Flow rate Double Conventional
2 A2i Ae
(L/min) cyclone cyclone
The static pressure difference, ( Pis Pes ), is measured by the
10 1.39 1.58
megnehelic gauge, and the dynamic pressure difference is calcu-
20 1.38 1.57
lated and added to ( Pis Pes ) to yield the total pressure drop. The
30 1.37 1.55
total pressure drops of the double cyclone and those of the con-
40 1.37 1.53
ventional cyclone at different ow rates are listed in Table 2. As
 A NOVEL DOUBLE CYCLONE 379

Table 2 addition, the regression line for the double cyclone is located to
Pressure drops (Pa) of the double cyclone and the conventional the left of those for conventional cyclones, indicating a better
test cyclone at different ow rates cyclone design in that the double cyclone allows lower pressure
drop or consumes less energy at the same normalized cut size.
Flow rate Double cyclone Conventional cyclone Thus it can be concluded that the double cyclone is indeed a more
(L/min) (Pascal) (Pascal) ef cient and less costly design than conventional cyclones.
10 12.5 5.4 The reason that the double cyclone is superior to the conven-
20 32.5 12.3 tional cyclones can be explained as follows. First, the additional
30 82.5 20.9 cylinder wall separates the cyclone body space into two annu-
40 145.0 45.2 lar sections. Recently, several researchers found that the annular
dimension value is the key radial design factor in determining
cyclone performance (Moore and McFarland 1993, 1995; Kenny
an additional vortex in the double cyclone ensures a higher col- and Gussman 1997; Liden and Gudmundsson 1997). Therefore
lection ef ciency, which will cause an increase in pressure drop. an additional cylinder wall separates one big annular section
Since pressure drop is basically a measure of the energy that in a conventional cyclone design into two smaller annular sec-
a cyclone consumes and the normalized cut size characterizes tions in the double cyclone design and may help to enhance
the effectiveness of a cyclone design, a correlation of these two the collection ef ciency. The added cylinder wall also serves
factors to examine their relationship should provide useful data. as a particle collection surface providing additional surface for
Figure 8 depicts the total pressure drop versus of the normalized particles to deposit. Second, due to the design, particle-laden
cut size for both the double cyclone and conventional cyclones air is forced to migrate an additional vortex inside the double
studied by previous researchers and in this study. A linear re- cyclone. Since centrifugal force is the main particle-removing
lationship is observed for most cyclone designs in Figure 8. In mechanism inside a cyclone, an additional vortex provides added

Figure 8. Pressure drop versus normalized cut size for the different cyclones.
380 Y. ZHU ET AL.
centrifugal force to collect more particles. Thus the development
of this additional vortex contributes signi cantly to particle col-
lection ef ciency. To fully develop the vortex, a higher ow
rate is preferable in that it provides stronger rotational motion
of the air. This may be the reason that the double cyclone pro-
vides increasingly higher collection ef ciency than the conven-
tional test cyclone when the air ow rate is increased as shown in
Figures 5, 6, and 7. Finally, the lower pressure drop inside the
double cyclone may be explained as follows. Due to the dou-
ble cyclone design, at the inlet into the second cyclone the air
stream does not enter through a single inlet port, but along the
entire perimeter of the middle cylinder. Thus the pressure drop
at the inlet to the second cyclone is reduced in relation to what
it should have been for an ordinary single-port tangential inlet
at the same diameter (23 mm).
CONCLUSION
The experimental results in this study provide convincing
evidence that the double cyclone is a superior design compared
to conventional cyclones. The double cyclone has a higher col-
lection ef ciency, a sharper separation curve, and consumes less
energy to achieve the same normalized cut size. It is believed that
all of these advantages are due to the added cylinder wall in the
double cyclone design that separates the limited cyclone body
space into two annular sections and forces the particle-laden air
to travel three vortexes before it leaves the cyclone.
REFERENCES
Boysan, F., Ayers, W. H., and Switherband, J. (1982). A Fundamental Mathemat-
ical Modelling Approach to Cyclone Design, Trans. ICHEME 60:222 230.
Coker, A. K. (1993). Understand Cyclone Design, Chem. Eng. Prog. 51 55.
Daniels, T. C. (1957). Investigation of a Vortex Air Cleaner, Engineering
(London) 8:258.
Dietz, P. W. (1981). Collection Ef ciency of Cyclones Separators, AICHE J.
27:888 892.
Dirgo, J., and Leith, D. (1985). Cyclone Collection Ef ciency: Comparison
of Experimental Results with Theoretical Predictions, Aerosol Sci. Technol.
4:401 411.
Grif ths, W. D., and Boysan, F. (1996). Computational Fluid Dynamics (CFD)
and Empirical Modelling of the Performance of a Number of Cyclone Sam-
plers, J. Aerosol Sci. 27/2:281 304.
Grinshpun, S. A., Willeke, K., Ulevicius, V., Juozaitis, A., Terzieva, S., Donnelly,
J., Stelna, G. N., and Brenner, K. P. (1997). Effect of Impaction, Bouncing,
and Reaerosolization on the Collection Ef ciency of Impingers, Aerosol Sci.
Technol. 26:326 342.
Ioza, D. L., and Leith, D. (1989). Effect of Cyclone Dimensions on Gas
Flow Pattern and Collection Ef ciency, Aerosol Sci. Technol. 10:491
500.
Ioza, D. L., and Leith, D. (1990). The Logistic Function and Cyclone Fractional
Ef ciency, Aerosol Sci. Technol. 12:598 606.
Jaroszczyk, T., and Ptak, T. (1985). Experimental Study of Aerosol Separation
using a Minicyclone, Powder and Bulksolid Confer. (10th Annual), pp. 611
622.
Kenny, L. C., and Gussman, R. A. (1995). Characterization and Modelling
of a Family of Cyclone Aerosol Preseparators, J. Aerosol Sci. 26:S777
S778.
Kim, J. C., and Lee, K. W. (1990). Experimental Study of Particle Collection by
Small Cyclones, Aerosol Sci. Technol. 12:1003 1015.
Lee, K. W., Gieseke, J. A., and Piispanen, W. H. (1985). Evaluation of Cyclones
Performance in Different Gases, J. Atmospheric Env. 19/6:847 852.
Li, E., and Wang, Y. (1989). A New Collection Theory of Cyclone Separators,
AICHEJ 35:666 669.
Liden, G., and Gudmundsson , A. (1997). Semi-Empirical Modelling to Gener-
alise the Dependenc e of Cyclone Collection Ef ciency on Operating Condi-
tions and Cyclone Design, J. Aerosol Sci. 28:853 874.
Moore, M. E., and McFarland, A. R. (1990). Design of Stairmand-Type Sampling
Cyclones, Am. Ind. Hyg. Assoc. 51/3:151 159.
Moore, M. E., and McFarland, A. R. (1993). Performance Modelling of Single-
Inlet Aerosol Sampling Cyclone, Environ. Sci. Technol. 27:1842 1849.
Moore, M. E., and McFarland, A. R. (1995). Design Methodology for Multiple
Inlet Cyclones, Environ. Sci. Technol. 30:271 276.
Ogawa, A. (1984). Cyclone Dust Collectors. In Separation of Particles from Air
and Gases, edited by J. Beddow. CRC Press, Inc., FL.
Razgatis, R., and Gusenther, D. A. (1981). Separation Ef ciency of a Cy-
clone Separation with Turbulence Suppressing Rotating-Insert, J. Eng. Power.
103:566 571.
Stairmand, C. J. (1951). The Design and Performance of Cyclone Separators,
Trans. Instn. Chem. Engrs. 29:356 383.
Ulevicius, V., Willeke, K., Grinshpun, S. A., Donnelly, J., Lin, X., and Mainelis,
G. (1997). Aerosolization of Particles from a Bubbling Liquid: Characteristics
and Generator Development , Aerosol Sci. Technol. 26:326 342.
Watson, J. G., and Chow, J. C. (1993). Ambient Air Sampling. In Aerosol Mea-
surement: Principles, Techniques and Applications., edited by K. Willeke and
P. A. Baron. Van Nostrand Reinhold, New York, pp. 622 639.
Willeke, K., Lin, X., and Grinshpun, S. A. (1998). Improved Aerosol Collec-
tion by Combined Impaction and Centrifugal Motion, Aerosol Sci. Technol.
28:439 456.
Zhu, Y., and Lee, K. W. (1999). Experimental Study on Small Cyclones Oper-
ating at High Flow Rates, J. Aerosol Sci. 30:1303 1315.